The proposed research evaluates a model of speech learning by examining the changes in speech production, speech perception, and speech processing that occur when individuals are exposed to the phones found in a foreign language (L2). One important hypothesis to be tested is that equivalence classification prevents individuals who learn an L2 after the age of 12;0 years (late L2 learners), but not those who learn an L2 before the age of 7;0 years (early L2 learners), from producing similar L2 phones authentically. It is hypothesized that early but not late L2 learners establish a phonetic category for similar L2 phones. As a result, early L2 learners with sufficient L2 experience will produce similar L2 phones authentically by implementing them with a different phonetic category than their counterparts in the native language (L1). Late L2 learners, on the other hand, do so by forming a phonetic realization rule with which to produce similar L2 phones. Equivalence classification should not prevent either early or late L2 learners from establishing a phonetic category for new L2 phones which have no counterpart in L1. It is hypothesized that early and late L2 learners develop different phonetic systems because equivalence classification differentially affects their use of the sensory information associated with similar L2 phones as a function of age of learning. These hypotheses are tested in an integrated series of 13 experiments which will examine spontaneous speech production, rapid switches in speech production between L1 and L2, imitation, speeded classification, similarity scaling, and AX discrimination. The research culminates in a training study that aims to precipitate the establishment of a phonetic category for a similar L2 phone through auditory or motor skill training.